Search

Small Cell Cloud (SCC) consists of Cloud-enabled Small Cells (CeSCs), which serve as radio end-points for mobile user equipments (UEs) and host computation offloaded from mobile UEs. SCC hereby brings advantages of a centralized cloud computation to the users' vicinity. The SCC architecture provides a mechanism for distribution of computation demand across the CeSCs. An effectiveness of the offloading is determined based on quality of radio channel between the UEs and the CeSC and predicted computation complexity. In this paper, we introduce an implementation of an offloading framework to facilitate adaptation of mobile apps for the SCC and to handle low-level communication between the app and the SCC. An evaluation of the offloading framework is conducted using Augmented Reality (AR) app, which requires intensive computations and low latency. The offloading framework and the AR app are a basement for the SCC testbed used to proof the concept of the computation offloading. Various computation and radio parameters are investigated to reveal benefits of the SCC. According to the performed measurements, the computation offloading can decrease latency up to 88 % and energy consumption of the UEs up to 93 %.

Current mobile devices evolve quickly and bring more powerful hardware to users. However, rate of such development surpasses advances in creation more durable batteries. Although energy awareness of mobile devices is utmost important, the significance of results of energy consumption modeling drops rapidly due to never-ending massive release of new mobile devices. In this paper, we describe new methodology and tool developed for measurement of energy consumption of user devices. Using this new tool, we carry out measurements of selected non-radio components (CPU, screen, I/O storage operations, speakers) of smartphones. For each component, several parameters are investigated. These parameters influence how a component is used by a user and therefore have great impact on overall energy consumption. Moreover, mathematical models are proposed in order to predict energy consumption without necessity to perform measurements and to determine expected consumption in real time.

Multimodal human-computer interaction combining multiple input modalities is the key for permitting our highly skilled and coordinated communicative behavior to control computer systems in flexible and natural manner. In this paper a multimodal interaction platform, called Manitou, is presented. The platform empowers researchers and developers to employ speech, gestures and other modalities in applications, enabling multimodal interaction. Key features of the platform are exposed to the web environment to facilitate a fast prototyping and delivering applications, exploiting multimodal interfaces. Benefits of the platform have been proven by a sample application and subsequent user experience survey.